OpenAI has built the next big door into personal data. From Friday, ChatGPT Pro subscribers in the US can connect their bank accounts to ChatGPT - through Plaid, with access to more than 12,000 financial institutions, including Chase, Schwab, Fidelity, Robinhood, American Express and Capital One.

What do users get? A dashboard with portfolio, spending, subscriptions and upcoming payments. They can ask ChatGPT things like "I feel like I'm spending more lately. Has something changed?" or work out a plan to buy a flat. ChatGPT then answers with access to their entire financial history.

The context matters. In April, OpenAI bought the team behind the startup Hiro, which was working on personal finance, and is now leaning on the new model - GPT-5.5 - for financial answers with improved contextual reasoning. The company claims that more than 200 million people a month already ask financial questions on ChatGPT - but most of those questions were theoretical, with no access to real data. Now those same users walk into a strange new zone with a single click.

The question everyone avoids: what happens to the data? OpenAI says users can unlink accounts and delete the data through Settings, with automatic deletion after 30 days. Sounds fine - until you remember that a few weeks ago the same company admitted that hackers had stolen some of the data in a security incident. For Balkan readers who still remember older banking hacks, this can look like signing up for free storage on a risk that's worth a lot more than a subscription fee.

The product is currently available only to Pro users in the US, on web and iOS, with a rollout to Plus subscribers after testing. Next up: integration with Intuit for tax-impact calculations and predictions of credit capacity. In other words - an algorithm that will know whether the bank will approve your mortgage before your bank knows the answer. How long before tools like this reach Europe?